CGD Research: CAS Overview

Climate Analysis Section

Diagnostic analyses of extratropical ocean-atmosphere-sea ice variability using models and observations

Using suites of atmospheric general circulation model (AGCM) simulations forced by global, regional, and idealized sea surface temperature (SST) variations Hoerling et al. (2006) found that multi-decadal variations and trends in SSTs have determined the spatial patterns, time history and seasonality of observed changes in African rainfall since 1950. The multi-model ensemble mean from 80 separate 50-yr Global Ocean Global Atmosphere (GOGA) simulations from five different AGCMs realistically captured the observed seasonality and spatial structure of downward trends in African rainfall, not only over the Sahel during boreal summer, but across the continent following the seasonal migration of rainfall, e.g., see Fig. 5. That the observed drying trend fell within the distribution function of the simulated trends indicates that it was likely a consequence of 20th Century SST variations. Moreover, no single 50-yr trend in unforced coupled ocean-atmosphere model simulations yielded a drying rate as large as observed (blue curve in Fig. 5), and nearly half of the GOGA rainfall trends fell outside the PDF from unforced coupled models, suggesting the responsible air-sea interactions were not arising from natural variations alone. The Indian Ocean warming thus appears to have been a key source for a decline in the austral summer African monsoon, though its role in forcing drought over the Sahel remains unclear.

A new monthly surface boundary forcing data set for uncoupled simulations with the CAM has been developed (Hurrell et al. 2006c). It is a merged product based on the monthly mean Hadley Centre sea ice and SST data set version 1 (HadISST1) and version 2 of the NOAA weekly optimum interpolation (OI.v2) SST analysis from 1870 to the present: it is updated monthly, and it is freely available for community use. The merging procedure was designed to take full advantage of the higher resolution SST information inherent in the OI.v2 analysis, which arguably is the best global SST analysis currently available.

The dynamical simulation of the latest version of the Community Atmosphere Model (CAM3) was examined as part of a special issue of J. Climate devoted to CCSM3 (Hurrell et al. 2006b). They used an ensemble of integrations forced with observed monthly varying SSTs and sea ice concentrations to examine the mean state and interannual variability of the CAM3, and to compare those features to coexisting observations (Fig. 6). The most significant differences from the previous version of the model (CCM3) were associated with changes to the parameterized physics package. Results showed that these changes have resulted in a modest improvement in the overall simulated climate; however, CAM3 continues to share many of the same biases exhibited by CCM3. The simulation of ENSO in CCSM3 was evaluated in the same special issue of J. Climate (Deser et al. 2006a). They found that the spatial patterns and amplitudes of ENSO are more realistic in CCSM3 than earlier versions of CCSM; however, the frequency of ENSO continues to be too short (2-3 years) compared to nature (3-8 years). The atmospheric teleconnections from the tropical Pacific to higher latitudes are improved in the high resolution version of CCSM3 (T85 spectral truncation) relative to the low resolution version (T42) and earlier versions of CCSM. Understanding the transient evolution of the atmospheric circulation response in CCM3 to imposed North Atlantic/Arctic SST and sea ice anomalies was also pursued through a 240-member ensemble of experiments to show how the response evolves (Deser et al., 2006b). In the first 2-3 weeks, the geopotential height response exhibits a baroclinic vertical structure localized to the vicinity of the forcing but evolves into an equivalent barotropic pattern that is hemispheric in scale and projects strongly onto the leading mode of internal variability in the control ensemble. This equilibrium response achieves maximum amplitude after 2-2.5 months. A linear baroclinic model used to diagnosis the processes involved in the two stages of the response, shows that the equilibrium response is maintained primarily by the transient eddy heat and vorticity flux convergences, while the initial response is maintained by anomalous diabatic heating associated with the imposed thermal forcing. An investigation of the impact of the SST ''re-emergence'' mechanism upon the winter-to-winter persistence of the NAO using an entraining ocean mixed layer model coupled to CAM2 (Cassou et al., 2006) shows that the reemergence of the SST "tripole" pattern in the North Atlantic enhances the winter-to-winter persistence of the NAO by 15-20%.

An analysis of North Pacific decadal variability in the 1000-yr control run of CCSM2 (Kwon and Deser, 2006) has shown that anomalous geostrophic advection and associated heat divergence in the upper 200m of the ocean forces decadal (~20 yr and ~40 yr) SST anomalies in the Kuroshio Current Extension, with surface heat fluxes and Ekman currents responding to rather than driving the SST anomalies. The anomalous geostrophic currents are a result of basin-scale wind stress curl anomalies 3-5 years earlier. These results suggest that the simulated North Pacific decadal variability in CCSM2 owes its existence to two-way ocean-atmosphere coupling.

A combined observational and modeling analysis of the extent to which the 1976-77 climate transition over the North Pacific was forced by the tropics was carried out using the TOGA AMIP ensemble integrations with CAM3 for the period 1950-2001 (Deser and Phillips 2006). Approximately 75% of the “shift” is attributable to tropical SST forcing in the 10-member CAM3 AMIP ensemble. However, a similar ensemble size of CCM3 AMIP integrations shows no appreciable response in the Aleutian Low. The differences between the two model responses were traced to differences in their precipitation responses over the tropical Indian Ocean and, based on a combination of observational constraints and dynamical theory, the large and erroneous increase in rainfall over the tropical Indian Ocean after 1977 simulated by CCM3 is responsible for that model's poor simulation of the 1976-77 climate shift of the Aleutian Low.

The temporal variability of ocean heat uptake in observations and in climate models was explored by AchutaRao et al. (2006). Given information about the distribution of observations in World Ocean Atlas WOA-2004, this study evaluated the effects of sparse observational coverage and the infilling used to produce the spatially-complete temperature fields required to compute heat content variations. Subsampling model data with actual observational coverage has a large impact on the inferred temperature variability in the top 300 and 3000 meters of the ocean. This arises from changes in both sampling depth and in the geographical areas sampled.

Unusual temperature trends in California (Bonfils et al. 2006) are evaluated with help of model runs. Observed increases in annual-mean surface temperature are not always distinguishable from climate noise, depending on the dataset considered. However, the large positive trends in mean and daytime temperatures in late winter/early spring as well as in nighttime temperatures from January to September are inconsistent with natural internal climate variability, and thus require one or more external forcing agents to be explained.

 

References

 

CGD Sectional Narratives

CAS | CCR | CDP | CMS | CSEG | OS | TSS |

 

 

 

 

 

 

Hydrological cycle

Figure 1. (High resolution image) The hydrological cycle. Estimates of the main water reservoirs, given in plain font in 103 km3, and the flow of moisture through the system, given in slant font in103 km3/yr, equivalent to Exagrams (1018 g) per year. (Trenberth et al. 2006a).

 

Hurricane Katrina Hours 42-54

Figure 2. (High resolution image) For 1800 UTC 28 August to 0600 UTC 29 August 2005, hours 42 to 54 of the simulation, given are (left) the azimuthally-averaged precipitation (mm/h), and (right) column integrated moisture convergence and surface latent heat flux as a function of radius for the control (red) and changes in SST of +1ºC (blue) and -1ºC (green). The precipitation and latent heat fluxes are area averages from the eye to the radius plotted to be compatible with the moisture convergence across that cylinder radius. (Trenberth et al. 2006b).

 

Yearly Surface Hurricane Flux

Figure 3. (High resolution image) Based on best track data for the tropical cyclones observed each year classified by maximum wind speed, the total surface energy loss by the global ocean is given based on Katrina simulated fluxes within 400 km of the eye of the storms as given by (2) as latent (blue), sensible (cyan) and total enthalpy (black) flux in 1021 Joules per year. Also given in green (right hand scale) is the precipitation in the same units. The dotted lines are linear trends. Trenberth et al (2006c).

 

Annual-mean frequency and intensity of daily precipitation

Figure 4. (High resolution image) Annual-mean frequency (% of time, left column) and intensity (mm/day, right column) of daily precipitation (>1 mm/day) events from TRMM satellite observations (top panels, 3B42 data set, 1998-2003 mean) and four different coupled models (1991-2000 mean). From Dai 2006a. Note the underestimates of intensity and overestimates of frequency of precipitation in the models.

 

1950-1999 trends

Figure 5. (High resolution image) The 1950-1999 trends in observed (left), GOGA (middle), and Indian Ocean forced (right) FMA rainfall (mm). The observed trend (gray bar) and PDFs of 50-yr rainfall trends averaged over the indicated region are shown in far right panel. The red (black) curve is from 80 (60) individual members of the GOGA (Indian Ocean) runs. The blue curve is from 15 members of unforced coupled model simulations.

 

ENSO event anomaly composites of precipitation

Figure 6. (High resolution image) Warm – cold ENSO event anomaly composites of precipitation during DJF for CAM3 (top) and CMAP (bottom). The contours are ? 0.25, 0.5, 1, 2, 4, and 8 mm day-1.